ABSTRACT
Associative learning is the acquisition of knowledge about the relationships between environmental features. The main objective of this study is to contribute to the ongoing debate that opposes associative and inferential theories of learning. According to inferential theories, learning is the result of inferences and reasoning on propositional representations. As a consequence, awareness of a rule always precedes the corresponding behavioral changes. According to associative theories, however, learning involves the gradual updating of the associative strength between the mental representations of the stimuli. On this view, consciousness is not mandatory for learning to take place. To test these contrasted predictions, we studied how behavior and consciousness change over the course of learning. Participants played a “Space Invaders” game in which they had to repeatedly press the spacebar to destroy Martian spaceships to prevent them from landing on Earth. Occasionally, they also had to avoid shooting at a protective Martian shield, which reflected back laser shots if fired upon, hence resulting in an immediate and massive invasion. Unbeknownst to participants, the shield’s occurrence was announced by cryptic signals (in Experiment 1, sounds and signs written in a Martian alphabet; in Experiment 2, sounds in Martian language), the signification of which was not conveyed to participants. In Experiment 1, during learning, participants also had to report, after each block, whether they believed their performance had improved and how they thought they had managed to improve it. In Experiment 2, participants had to help a control tower decide whether the shield would appear or not. We obtained measures of both the behavioral manifestation of associative learning (by recording bar-pressing rates before the occurrence of the shield) and its availability to awareness. Results suggest that participants learn the task in a gradual manner, as bar-pressing rate decreased after the presentation of the predicting cryptic signal. In Experiment 1, analysis of the verbal reports indicates that although some participants use intentional strategies and explicit rules to perform the task, some also exhibit learning of the association through their behavioral responses while remaining unable to describe the rule verbally. In Experiment 2, we observed that attention and the quality of representation between two events have an effect on learning: learning becomes conscious faster if participants learn intentionally and if the associated events are actually presented. In closing, we discuss the extent to which these results constitute evidence of unconscious associative learning and whether they are congruent with a dual-process learning theory.
